{
  "id": 239427,
  "title": "How to use test metadata",
  "url": "/competitions/birdclef-2021/discussion/239427",
  "author_name": "",
  "post_date": "2021-05-16T10:02:30.077343Z",
  "votes": 9,
  "comment_count": 7,
  "views": 0,
  "content": "<p>Hello, I've trained a model using date, latitude, and longitude, but I have a hard time figuring how to retrieve these metadata for the hidden testing dataset since it's hidden.<br>\nI see in the test_soundscapes folder 4 .txt files for location, but they don't refer to any file id, just to a site (COR, COL, SNE, SSW). Does this mean that all tests recording for a specific website were done at the same place? That's for latitude and longitude.<br>\nHow can we get the date of the recording as well? Is this included in the file's name?</p>",
  "messages": [
    {
      "id": "1309797",
      "postDate": "05/16/2021 10:02:30",
      "content": "<p>Hello, I've trained a model using date, latitude, and longitude, but I have a hard time figuring how to retrieve these metadata for the hidden testing dataset since it's hidden.<br>\nI see in the test_soundscapes folder 4 .txt files for location, but they don't refer to any file id, just to a site (COR, COL, SNE, SSW). Does this mean that all tests recording for a specific website were done at the same place? That's for latitude and longitude.<br>\nHow can we get the date of the recording as well? Is this included in the file's name?</p>",
      "rawMarkdown": "Hello, I've trained a model using date, latitude, and longitude, but I have a hard time figuring how to retrieve these metadata for the hidden testing dataset since it's hidden.\nI see in the test_soundscapes folder 4 .txt files for location, but they don't refer to any file id, just to a site (COR, COL, SNE, SSW). Does this mean that all tests recording for a specific website were done at the same place? That's for latitude and longitude.\nHow can we get the date of the recording as well? Is this included in the file's name?",
      "votes": null
    },
    {
      "id": "1309809",
      "postDate": "05/16/2021 10:12:19",
      "content": "<p>I've read again the kaggle data description and for the date it is said that \"The file names include the date the recording was taken, which can be especially useful for identifying migratory birds\" for the hidden test dataset. So for the date it is solved but not for latitude and longitude.</p>",
      "rawMarkdown": "I've read again the kaggle data description and for the date it is said that \"The file names include the date the recording was taken, which can be especially useful for identifying migratory birds\" for the hidden test dataset. So for the date it is solved but not for latitude and longitude.",
      "votes": null
    },
    {
      "id": "1309862",
      "postDate": "05/16/2021 11:03:08",
      "content": "<ul>\n<li>Latitude and longitude for the approximate location of all four sites are given to us in the text files. They aren't precise locations as far as I'm aware, but in the context of creating subsets of birds which are more likely to occur at these four sites, it seems sufficient  </li>\n<li>All soundscape recordings hidden or non hidden are tied to these locations.</li>\n</ul>\n<p>Hopefully someone can correct me if I'm wrong :)</p>",
      "rawMarkdown": "Latitude and longitude for the approximate location of all four sites are given to us in the text files. They aren't precise locations as far as I'm aware, but in the context of creating subsets of birds which are more likely to occur at these four sites, it seems sufficient  \n- All soundscape recordings hidden or non hidden are tied to these locations.\n\nHopefully someone can correct me if I'm wrong :)",
      "votes": null
    },
    {
      "id": "1309869",
      "postDate": "05/16/2021 11:09:47",
      "content": "<p>Thank you very much!</p>",
      "rawMarkdown": "Thank you very much!",
      "votes": null
    },
    {
      "id": "1310106",
      "postDate": "05/16/2021 13:45:13",
      "content": "<p>Isn't the site part of soundscape filename?</p>\n<p>Edit: LOL, answered hours after this was already answered, sorry.</p>",
      "rawMarkdown": "Isn't the site part of soundscape filename?\n\nEdit: LOL, answered hours after this was already answered, sorry.",
      "votes": null
    },
    {
      "id": "1310437",
      "postDate": "05/16/2021 16:53:26",
      "content": "<p>I have embedded month, rounded latitude, rounded longitude and concatenated it to the conv results : it doesn't seem to improve anything.</p>",
      "rawMarkdown": "I have embedded month, rounded latitude, rounded longitude and concatenated it to the conv results : it doesn't seem to improve anything.",
      "votes": null
    },
    {
      "id": "1330729",
      "postDate": "06/01/2021 04:29:10",
      "content": "<p>I'm still confused by this too.  Are we expected to know the format of the test soundscape file names?   Otherwise how can we extract the date from them.   The location question is answered, but I'm still struggling to see how to get the dates matched to the test soundscape.</p>\n<p>There are dates listed in    test_set_recording_dates.csv    But they don't uniquely match any particular recording, just the location.</p>",
      "rawMarkdown": "I'm still confused by this too.  Are we expected to know the format of the test soundscape file names?   Otherwise how can we extract the date from them.   The location question is answered, but I'm still struggling to see how to get the dates matched to the test soundscape.\n\nThere are dates listed in    test_set_recording_dates.csv    But they don't uniquely match any particular recording, just the location.",
      "votes": null
    },
    {
      "id": "1330887",
      "postDate": "06/01/2021 06:38:35",
      "content": "<p>I've found the answer <a href=\"https://www.kaggle.com/stefankahl/birdclef2021-exploring-the-data\" target=\"_blank\">here</a><br>\nThe files are in the format “1234_SSW_20170101.ogg”  the same as the train soundscapes.   Somehow that just wasn't obvious to me.</p>",
      "rawMarkdown": "I've found the answer [here](https://www.kaggle.com/stefankahl/birdclef2021-exploring-the-data)\nThe files are in the format “1234_SSW_20170101.ogg”  the same as the train soundscapes.   Somehow that just wasn't obvious to me.",
      "votes": null
    }
  ],
  "comments": [
    {
      "id": 1309809,
      "author_name": "josephamigo",
      "author_url": "",
      "post_date": "05/16/2021 10:12:19",
      "content": "<p>I've read again the kaggle data description and for the date it is said that \"The file names include the date the recording was taken, which can be especially useful for identifying migratory birds\" for the hidden test dataset. So for the date it is solved but not for latitude and longitude.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1309862,
          "author_name": "shtrausslearning",
          "author_url": "",
          "post_date": "05/16/2021 11:03:08",
          "content": "<ul>\n<li>Latitude and longitude for the approximate location of all four sites are given to us in the text files. They aren't precise locations as far as I'm aware, but in the context of creating subsets of birds which are more likely to occur at these four sites, it seems sufficient  </li>\n<li>All soundscape recordings hidden or non hidden are tied to these locations.</li>\n</ul>\n<p>Hopefully someone can correct me if I'm wrong :)</p>",
          "votes": null,
          "replies": []
        },
        {
          "id": 1309869,
          "author_name": "josephamigo",
          "author_url": "",
          "post_date": "05/16/2021 11:09:47",
          "content": "<p>Thank you very much!</p>",
          "votes": null,
          "replies": []
        }
      ]
    },
    {
      "id": 1310106,
      "author_name": "cpmpml",
      "author_url": "",
      "post_date": "05/16/2021 13:45:13",
      "content": "<p>Isn't the site part of soundscape filename?</p>\n<p>Edit: LOL, answered hours after this was already answered, sorry.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1310437,
      "author_name": "josephamigo",
      "author_url": "",
      "post_date": "05/16/2021 16:53:26",
      "content": "<p>I have embedded month, rounded latitude, rounded longitude and concatenated it to the conv results : it doesn't seem to improve anything.</p>",
      "votes": null,
      "replies": []
    },
    {
      "id": 1330729,
      "author_name": "ollypowell",
      "author_url": "",
      "post_date": "06/01/2021 04:29:10",
      "content": "<p>I'm still confused by this too.  Are we expected to know the format of the test soundscape file names?   Otherwise how can we extract the date from them.   The location question is answered, but I'm still struggling to see how to get the dates matched to the test soundscape.</p>\n<p>There are dates listed in    test_set_recording_dates.csv    But they don't uniquely match any particular recording, just the location.</p>",
      "votes": null,
      "replies": [
        {
          "id": 1330887,
          "author_name": "ollypowell",
          "author_url": "",
          "post_date": "06/01/2021 06:38:35",
          "content": "<p>I've found the answer <a href=\"https://www.kaggle.com/stefankahl/birdclef2021-exploring-the-data\" target=\"_blank\">here</a><br>\nThe files are in the format “1234_SSW_20170101.ogg”  the same as the train soundscapes.   Somehow that just wasn't obvious to me.</p>",
          "votes": null,
          "replies": []
        }
      ]
    }
  ],
  "raw_markdown_by_id": {
    "1309797": "Hello, I've trained a model using date, latitude, and longitude, but I have a hard time figuring how to retrieve these metadata for the hidden testing dataset since it's hidden.\nI see in the test_soundscapes folder 4 .txt files for location, but they don't refer to any file id, just to a site (COR, COL, SNE, SSW). Does this mean that all tests recording for a specific website were done at the same place? That's for latitude and longitude.\nHow can we get the date of the recording as well? Is this included in the file's name?",
    "1309809": "I've read again the kaggle data description and for the date it is said that \"The file names include the date the recording was taken, which can be especially useful for identifying migratory birds\" for the hidden test dataset. So for the date it is solved but not for latitude and longitude.",
    "1309862": "Latitude and longitude for the approximate location of all four sites are given to us in the text files. They aren't precise locations as far as I'm aware, but in the context of creating subsets of birds which are more likely to occur at these four sites, it seems sufficient  \n- All soundscape recordings hidden or non hidden are tied to these locations.\n\nHopefully someone can correct me if I'm wrong :)",
    "1309869": "Thank you very much!",
    "1310106": "Isn't the site part of soundscape filename?\n\nEdit: LOL, answered hours after this was already answered, sorry.",
    "1310437": "I have embedded month, rounded latitude, rounded longitude and concatenated it to the conv results : it doesn't seem to improve anything.",
    "1330729": "I'm still confused by this too.  Are we expected to know the format of the test soundscape file names?   Otherwise how can we extract the date from them.   The location question is answered, but I'm still struggling to see how to get the dates matched to the test soundscape.\n\nThere are dates listed in    test_set_recording_dates.csv    But they don't uniquely match any particular recording, just the location.",
    "1330887": "I've found the answer [here](https://www.kaggle.com/stefankahl/birdclef2021-exploring-the-data)\nThe files are in the format “1234_SSW_20170101.ogg”  the same as the train soundscapes.   Somehow that just wasn't obvious to me."
  },
  "source": "meta"
}